It is the concept developed by Ralph Kimball. Dimensional data modeling is the technique to optimize the data storage in the dimensional data warehouse. Such an optimization is focused to the querying performance instead of volume of the data storage. So, it is known as a de-normalized data modeling technique to enable fast data query.
Two type of entities are consisted in the data model. They are the Fact and Dimension tables which represents the measures and the context of the business process event respectively. They are identified from analyzing the key business processes and organized in an appropriated grain.
The Dimensional Data Model forms a Star Schema or Snowflake Schema, with dimensions surrounding the fact table.
Best Practices
We may review the BI/DW Best Practices related to dimensional data modeling below:
- Best Practice #5 - Consolidate entities while keep correct grain
- Best Practice #6 - Denormalize the data in the user access layer
- Best Practice #7 - Atomic grain base fact table
- Best Practice #8 - Use Surrogate key
- Best Practice #9 - Contain date dimension in all fact tables
- Best Practice #10 - Handle unknown and null
- Best Practice #12 - Manage data integrity
- Best Practice #13 - Keep measures additive if possible
No comments:
Post a Comment